Affordable Access

Publisher Website

‘Instant’ cusums from a discrete analyser

Hindawi Publishing Corporation
Publication Date
DOI: 10.1155/s1463924685000104
  • Research Article
  • Chemistry


Journal of Automatic Chemistry, Volume 7, Number (January-March 1985), pages 46-48 ’Instant’ cusums from a discrete analyser P. Henry and D. C. Ephraim Medical Biochemistry Department, Llandough Hospital, CardiffCF61XX, UK Introduction Quality control is an important feature of today’s clinical chemistry laboratory. Most samples are processed in batches, which include quality-control samples. The values obtained for these quality controls can be used in various ways. The simplest way is to compare the results with the limits of acceptability. Another method is to plot the results graphically: this is more laborious, but it yields more information from the same data, in that it can reveal trends. Judging by the quality-control programs offered by several manufacturers of control sera, this is the way used by many laboratories. A further option is to plot the cumulative sum (cusum) of the results. This is more sensitive still, but is not widely used, probably because of the extra calculations needed, and because few of the standard textbooks on clinical chemistry include any directions on the use of cusums. Edwards [1] wrote a very helpful practical introduction to the use ofcusums, and with microcomputers now being available in many laboratories, extra calculations are no longer a problem. This paper proposes the use ofcusums, because more can be deduced about the assays being carried out by using them than can be deduced from a simple graphical plot of the same data. Theory The British Standards Institution’s recommendations on the use of cusums [2 and 3] have been followed in both symbols and terminology. In any method of assessing quality, two measures are needed to define the desired quality: the target value and the variation. When using cusums in clinical chemistry, where individual observations are made from a process under statistical control, the most suitable measure for the target value (T) is the mean, and for the variation, the standard error (o) calculated from the difference

There are no comments yet on this publication. Be the first to share your thoughts.


Seen <100 times